Advances in taxation by john handerlinde

prices at greater than 100%, suggesting prior underestimations based onmeasurement errors in public service variables.Keywords: Property taxes; public service benefits; quality of life;

Series Editor: John Hasseldine

Recent Volumes:

Volumes 1
3: Edited by Sally M Jones

Volumes 4 and 5: Edited by Jerold J Stern

Volumes 6
16: Edited by Thomas M PorcanoVolume 17 and 18: Edited by Suzanne LuttmanVolumes 19
21: Edited by Toby Stock

Volume 22: Edited by John Hasseldine

ADVANCES IN TAXATION

EDITED BY JOHN HASSELDINE Paul College of Business and Economics, Department of Accounting and Finance, University of New Hampshire,

Durham, NH, USA

India
Malaysia
China

First edition 2017

Copyright r 2017 Emerald Group Publishing Limited

Reprints and permissions service

Contact: permissions@emeraldinsight.com

No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center Any opinions expressed in the chapters are those of the authors Whilst Emerald makes every effort to ensure the quality and accuracy of its content, Emerald makes no representation implied or otherwise, as to the chapters’ suitability and application and disclaims any warranties, express or implied, to their use.

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

LIST OF CONTRIBUTORS vii

INTRODUCTION: PUBLISHING QUALITY

THE EFFECTS OF PROPERTY TAXES AND PUBLIC

SERVICE BENEFITS ON HOUSING VALUES: A

COUNTY-LEVEL ANALYSIS

MEASURING AND CHARACTERIZING THE DOMESTIC EFFECTIVE TAX RATE OF US CORPORATIONS

TAX AND PERFORMANCE MEASUREMENT: AN

INSIDE STORY

THE IMPACT OF CULTURE AND ECONOMIC

STRUCTURE ON TAX MORALE AND TAX EVASION:

A COUNTRY-LEVEL ANALYSIS USING SEM

THE DETERMINANTS OF TAX MORALE AND TAX

COMPLIANCE: EVIDENCE FROM JORDAN

v

MEASURING TAX COMPLIANCE ATTITUDES: WHAT SURVEYS CAN TELL US ABOUT TAX

COMPLIANCE BEHAVIOUR

Fadi Alasfour Al-Ahliyya Amman University, Jordan

Galway, Ireland

Galla

Salganik-Shoshan

Ben-Gurion University of the Negev, Israel

vii

John Hasseldine, Editor

University of New Hampshire, USA

Erich KirchlerUniversity of Vienna, AustriaStephen Liedtka

Villanova University, USAAlan MacnaughtonUniversity of Waterloo, CanadaAmin Mawani

York University, CanadaJanet A Meade

University of Houston, USAEmer Mulligan

National University of IrelandGalway, Ireland

Lynne OatsUniversity of Exeter, UKGrant RichardsonUniversity of Adelaide, AustraliaRobert Ricketts

Texas Tech University, USA

ix

University of Muenster, Germany

QUALITY TAX RESEARCH

As signaled in Volume 22 one of my goals is for Advances in Taxation tohave a greater international exposure This means carrying more articleswith international implications, authored from any country However, it iscritical that we continue the tradition of publishing high quality taxresearch To this end, I reiterate that Advances in Taxation will continue topublish, quality North-American tax research and that from otherjurisdictions providing it is of broad interest to our readers

I wish to thank the editorial board for their continued support They havebeen called upon to promote AIT and to engage in the reviewing process.Many have again provided wise counsel for this volume Apart from theeditorial board, I am also pleased to thank the ad hoc reviewers listed belowfor their valuable and timely reviewing activity during 2015
2016

May Bao (University of New Hampshire)

Jonathan Farrar (Ryerson University)

Brian Huels (Rockford University)

Teresa Lang (Auburn University at Montgomery)

Nor Aziah Abd Manaf (Universiti Utara Malaysia)

Mohd Rizal Palil (Universiti Kebangsaan Malaysia)

Jeff Pope (Curtin University)

Donna Bobek Schmitt (University of South Carolina)

Tanya Tang (Brock University)

Recep Yucedogru (University of Nottingham)

In this volume, there are six papers In the lead paper, Kimberly Key,Teresa Lightner, and Bing Luo extend literature in the property tax area,especially in helping to define and operationalize Quality of Life measuresthat explain property values Using composite rankings to measure theeconomy, education, health, and public safety, they provide evidence onhow property taxes are capitalized into housing prices Their study willhelp future researchers to more fully consider public service benefits in theirtax capitalization models

xi

Shifting from the micro-aspect of tax capitalization models, the secondpaper in this volume provides macro evidence on the domestic effective taxrate (ETR) of US corporations over the time period 2003
2010 YaronLahav and Galla Salganik-Shoshan investigate how domestic ETRs areaffected by factors representing business and financial structure along withmacroeconomic conditions While they acknowledge some of their findingsmight be anticipated, other results suggest the need for more research.Adopting a different methodological approach, but still focused in part onETRs, Emer Mulligan and Lynne Oats report on the findings of 26 semi-structured interviews conducted with tax executives from 15 Silicon Valleycorporations This study highlights the value of qualitative research as inter-viewing tax professionals allowed the authors to drill down and understandhow performance measures are used in tax departments and how tax as ameasure of organizational performance is presented to external stakeholders.The next three papers in this volume are an integrated forum on taxmorale and the measurement of compliance attitudes In the first paper ofthis forum, William D Brink and Thomas M Porcano use structural equa-tion modeling to develop a comprehensive international tax evasion frame-work by analyzing direct and indirect paths between country-level culturaland economic structural variables and tax morale and evasion.

In the second paper of the forum, Fadi Alasfour, Martin Samy, andRoberta Bampton review the literature on tax morale and issue a surveyinstrument to Jordanian tax auditors and financial managers Apart fromthe specific empirical results, this Jordanian study is notable as there is littleprior research on tax morale in non-Western countries and also for theirdevelopment of a multi-item index comprising 17 questions to measure par-ticipants’ “intrinsic motivations” to pay taxes

Finally, in a related methodological paper, Diana Onu thoughtfully ines the way that prior literature has researched the link between tax attitudemeasures and actual compliance behavior She suggests several avenues toimproving the predictive value of attitude measures and offers a number ofrecommendations that will prove useful to behavioral tax researchers

exam-In future volumes, I wish to signal that apart from continuing its tradition

of publishing original research-based manuscripts, Advances in Taxation willconsider publishing papers on methodological issues (as several of the papers

in this volume attest) and quality and topics papers on aspects of tax tion, the tax profession, and also well-crafted replications co-authored bydoctoral students and faculty

educa-John Hasseldine

Editor

TAXES AND PUBLIC SERVICE

Advances in Taxation, Volume 23, 1
31

Copyright r 2017 by Emerald Group Publishing Limited

All rights of reproduction in any form reserved

ISSN: 1058-7497/doi: 10.1108/S1058-749720160000023007

1

prices at greater than 100%, suggesting prior underestimations based onmeasurement errors in public service variables.

Keywords: Property taxes; public service benefits; quality of life; taxcapitalization

INTRODUCTIONThis study investigates the effects of local property taxes and public services

on residential property Property tax theory predicts that property valueswill be negatively related to the property taxes of a taxing jurisdiction, andpositively related to its public services This study is motivated by the lack

of theoretically appropriate public service measures in prior property taxresearch We examine four areas intended to capture broad quality-of-life(QOL) aspects of local jurisdictions
the economy, education, health, andpublic safety
and develop public service outcome measures for each Thisstudy is also motivated by the ongoing debate over the extent to whichproperty taxes are capitalized into property values Tax capitalization issuesare important because they pertain to the economic incidence of a tax, that

is, who bears the burden of the tax In full capitalization, the house valuewould include the expected tax liabilities, and the current owners wouldbear the entire burden of contemporary changes in expected tax liabilities.The economic significance of property taxes and public service delivery inlocal government decision-making is a third motivation for this work

By using primarily public service output measures (i.e., the QOL sures) rather than input measures (i.e., spending), this study overcomes one

mea-of the most significant problems in prior property tax research
the quate modeling of local public services.Palmon and Smith (1998) explainthat a downward bias exists in the tax capitalization coefficient, created byerrors in the measurement of public services and by the inherent relationbetween public services and tax rates This coefficient bias makes it difficult

inade-to isolate tax effects from public service effects in empirical property taxanalyses, resulting in an underestimation of the degree of tax capitalization.Most prior research relies on the assumption that higher spending improvesquality (Fischel, Oates, & Youngman, 2011); yet spending has long beenrecognized as a poor measure of quality (e.g., Oates, 1969; Rosen &

in large part has not sufficiently explored models that could explain theconnections between local determinations of public service distribution andthe housing markets They state that far more work needs to be done toobtain comprehensive measures of benefits and to estimate the impliedmarginal benefits This study addresses both with the QOL measures Afinal motivation for the study is to introduce tax considerations to socialscience QOL research The social sciences literature includes some variablesthat represent QOL, but they do not incorporate taxes.

tax capitalization and empirical research, and state that most studies find asignificant negative relation between property values and property taxes,but that the degree of capitalization in those studies has varied The mosttypical empirical result, they note in their summary of prior research, hasbeen the partial capitalization of property taxes; however, results rangefrom no significant capitalization to full capitalization and, in one case, towhat they describe as overcapitalization Overall, this study provides addi-tional evidence on tax capitalization and incidence questions, an importantarea of public finance research that is not well understood and that lacksconsensus (Fischel et al., 2011;Zodrow, 2001)

This study also addresses an economically large and important tax.Property taxes account for approximately 75% of local governmenttax revenue, and residential property accounts for approximately 60% oftaxable assessments, the largest component of the tax base by a significantmargin (Lutz, Molloy, & Shan, 2011) Local officials levy these propertytaxes and in turn determine allocations for public services to residents.Ultimately, they make tax and spending decisions targeted toward theQOL outcomes they believe are desirable in their jurisdictions This studyprovides evidence on whether local residents capitalize their taxes and QOLmeasures in the form of property values

We test the effects of property taxes and public service benefits onhousing values using data from 1999 to 2009 for 159 counties in Georgia.Property value measures are calculated using assessed values of propertygrossed up to the fair market value (FMV); local taxes include all localtaxes paid on that property (county, school, and city taxes) The localtaxes and property values are used to construct an effective tax rate(ETR) measure Each of the four QOL measures comprises subcompo-nents that help produce a broad measure of the category; they areadopted from a U.S county-level public policy analysis edited by Vocino

make up the four QOL measures; these subcomponent rankings are then

totaled to grade the county’s overall ranking for each QOL measure TheETR is predicted to be negatively related to property values, and theQOL variables are predicted to be positively related to property values.

We find a strong negative relation between county ETRs and tial property values This result is consistent with capitalization oflocal residential property taxes into housing values There is also a statisti-cally significant and positive association between three of the four QOLrankings
economy, education, and public safety
and Georgia residen-tial property values However, the health of a county was not significantlyrelated to housing values in Georgia counties Several control variablesare statistically significant as well Accordingly, we find that residentialproperty values reflect local property taxes, QOL measures, and socioeco-nomic factors The statistically negative relation between property taxesand housing values indicates some level of capitalization exists Given thestatistically negative relation for property taxes, this study makes animportant contribution in its estimation of the extent of tax capitalization.This result validates concerns about the underestimation of capitalization

residen-in prior research Full capitalization is consistent with the theory thathousing market participants rationally discount properties subject tohigher taxes, implying that only unexpected tax changes can be passed on

to new buyers of residential real estate (Palmon & Smith, 1998) A pointestimate of our data indicates greater than full capitalization

Prior tax capitalization research has not fully examined the marginalbenefits of public services While our regression results indicate that abetter economic environment and higher overall rankings in education andpublic safety are associated with higher housing values, we want to betterunderstand each factor’s marginal effect on those values We use t-tests ofstandardized coefficients to determine which QOL factors have the greatestimpacts As expected, the economy has the greatest influence on housingvalues on a statewide basis Next we divide the state into 12 regions andfind that the marginal effects of QOL factors differ by region In mostregions, economics affect housing prices more so than the other factors,but education is a close second Surprisingly, health factors have a greaterinfluence on housing values in more regions than does public safety.Overall, our results should encourage researchers in the property tax inci-dence area to fully consider public service benefits in their tax capitalizationmodels In addition, local government officials can also benefit from ourevidence Our findings have implications for the competitive environmentthat local policymakers face in attracting residents through wise tax andspending decisions on local public services

The remainder of this paper is organized as follows In the next section,

we discuss the motivation for the research and the prior literature on erty tax capitalization We also state the research hypotheses We proceed

prop-to review the QOL literature and explain the QOL measurements for thisstudy The Georgia property tax system is described next, including datasources The research design, results, and conclusions follow

MOTIVATION AND PRIOR RESEARCH

capitalization He adopts theTiebout (1956) view of the consumer ping” among communities that offer different tax-expenditure packages.Empirically, Oates regresses house prices on a vector of housing character-istics (the cost of taxes) and a public service measure (education expendi-tures per pupil) He finds a significant negative relation between propertyvalues and property tax rates, with about two-thirds capitalization Therelation between property values and expenditures is positive.Oates (1969)

“shop-states that the results are consistent with theTiebout (1956)model in whichpeople appear to be willing to pay more to live in a community that pro-vides higher levels of public services.1

values should be lower in communities with higher tax rates and average public services However, they argue that the Oates model is defi-cient because it proxies public service output with input expenditures.Instead of expenditures per pupil, they use a school achievement score.They employ the same 1960 data that Oates (1969) did, and add

below-1970 data They find tax capitalization rates are higher when the ment scores are used instead of expenditures, suggesting that bettermodel specification affects inferences about property taxes For 1970, theexpenditure-per-pupil is statistically insignificant, which shows that theexpenditures and performance measures are not capturing quite the sameunderlying construct

achieve-We follow Oates (1969), Rosen and Fullerton (1977), and other priorresearch to investigate the following research questions (stated inresearch form):

H1 There is a negative relation between residential property valuesand local property taxes

H2 There is a positive relation between residential property valuesand local public services.

While studies subsequent to Oates (1969) and Rosen and Fullerton(1977) have included from one to a few public service variables, none hasextensively examined the effects of local QOL indicators on residentialproperty values Since the earliest research, it has been recognized that pub-lic goods and services are difficult to measure, and that spending is a poormeasure of quality (e.g., Lewis & McNutt, 1979; Ross & Fullerton, 1977;

in the area are familiar with the difficulties in obtaining operational sures of output in the public sector.Palmon and Smith (1998)state that aninability to control adequately for public services creates an under-identifi-cation problem in tax capitalization models, resulting in lower estimates oftax capitalization rates

mea-Despite widespread recognition of these measurement issues, empiricalresearch to date has shown little improvement in overcoming them, and tothe extent that there are improvements, nearly all the public service vari-ables are education-related (e.g., student test scores).Oates (1969)uses cur-rent expenditures per pupil and municipal spending on all functions otherthan schools, and debt as his proxy for benefits;Hamilton (1976) employsper-household expenditure on local public services; McDougall (1976)

includes more variables and controls for benefits with grade-12 medianscore on the Iowa Tests for Educational Development, and includes as wellvariables that measure crime rate, the number of subfunctions of the parksand recreation services, and a fire department variable Ross and Yinger

comprehen-sive measures of benefits and to estimate the implied marginal benefits.This study addresses that call for more research It employs 18 indicatorsthat measure different aspects of public services and community QOL Werefer to these as (QOL) variables, consistent with related social scienceresearch As stated in Hypothesis 2, the four broad QOL measures arepredicted to be positively related to residential property values Prior QOLresearch and this study’s measurement of the QOL variables are discussed

in the next section

This study improves the residential property valuation model and allowsfor better inferences regarding the extent of property tax capitalization ifdata are consistent with Hypothesis 1, that is, property value and propertytax relation, is statistically negative.Sirmans et al (2008)review 28 tax capi-talization papers and find ten studies with partial capitalization, nine with

full capitalization, one with overcapitalization, and seven with no significantcapitalization.2 They conclude that the most typical empirical result hasbeen partial capitalization Nonetheless, results for the extent of capitaliza-tion in prior research are at all levels; no true consensus exists Our studyprovides new evidence on this important public finance issue, using a modelthat is better specified.

QOL RESEARCH AND MEASURES

The provision of public services that maintain and improve the QOL for ajurisdiction’s residents is one of the implicit mandates of modern govern-ment Several prior studies have examined how various standard-of-livingvariables affect QOL These measures range from fertility, health, and theenvironment to consumption, economics, migration, and individual rights,among others.3 These studies range in scope from international to inter-county and inter-city None of the studies has explicitly included a tax vari-able in its QOL models or indexes

We construct QOL measures based onVocino (2011), who uses variousindicator variables to form QOL factors to assess the performance of allcounties in Alabama The indicators include growth, public safety, andwell-being, along with poverty and income measures; but again there are

no measures of taxation or residential property values The variables ture aspects of county residents’ lives that affect their QOL
and thatlocal governments can alter through the provision of public services Thisstudy uses 18 indicators to quantify four QOL factors within a county:economy, education, health, and public safety.4

study, including the 18 indicators used to derive the four QOL variables

In order to standardize and combine the information into the QOL ables, we first rank each county from 1 to 159, worst to best, respec-tively, on the indicators that compose each QOL variable Ranking thebest county highest rather than number 1 improves the interpretation ofregression results Next we combine the individual indicator rankings toderive a composite score that is used to determine a county’s ranking foreach of the four QOL variables

vari-For example, on the education QOL factor for 1999, out of 159 counties,Appling County ranks 92nd on percentage of the population lacking basicliteracy skills, 66th on high school dropout rate, 35th on teacher
studentratio, 10th best on education funding per student, and 136th best on

Table 1 Description and Data Sources for All Variables.

Dependent Variable

RESVALUE ij Log of the assessed value of

residential property for each county for each year, 1999
2009, divided by 0.4

Georgia County Guide

Independent Variables

ETR ij The effective property tax rate for

each county, calculated as total local property taxes paid/(assessed value of property/0.4)

Georgia County Guide

ECrnk ij Income per capita Georgia County Guide

An average of each county’s

annual ranking on the

following economic factors for:

Annual unemployment rate Georgia County Guide Poverty rate Georgia County Guide Average weekly wage Georgia County Guide EDrnk ij Percentage of population lacking

basic literacy skills

National Center for Education Statistics

http://nces.ed.gov/naal/ estimates/

StateEstimates.aspx

An average of each county’s

annual ranking on the

following education factors:

High school dropout rate Georgia County Guide Teacher-student ratio Georgia County Guide Education funding per student Georgia County Guide Percentage of population with a

bachelor’s degree or higher

Georgia County Guide HCrnk ij Life expectancy 2006 Partner Up! for Public

Health http://www togetherwecandobetter com/allcountiesdb.html

An average of each county’s

annual ranking on the

following health factors:

Infant mortality rate Georgia County Guide Percentage of

rape, robbery, and aggravated assaults)

Georgia County Guide

Table 1 (Continued)

An average of each county’s

annual ranking on the

following public safety factors:

Property crimes reported (burglary, larceny, motor vehicle thefts)

Georgia County Guide

Juvenile arrests Georgia County Guide Adult arrests Georgia County Guide SALESTX ij Sales tax rate for each county Georgia County Guide lATL j Log of the distance from the

county seat to the Atlanta airport

MapQuest lPOP ij Log of total population of

University of Georgia,

http://spock.fcs.uga edu/hace/gafacts / AGE65 ij The percentage of the county

population age 65 or older

Georgia County Guide AGE018 ij The percentage of the county

population age birth to 18

Georgia County Guide INDDIST j 1 for the following counties:

Fulton, Haralson, Gwinnett, Gordon, Carroll, Bartow, Walker, Jackson, Whitfield, Dekalb, Laurens, Hall, Cobb, Mitchell, Floyd, Walton, Thomas, Chattooga, Toombs, and Lowndes; and 0 otherwise

Georgia County Guide

CPI i Annual average consumer

price index

U.S Bureau of Labor Statistics BUSVALUE ij The per capita assessed value of

commercial property in a county

Georgia County Guide REGION j A dummy variable for each

region, 1
11

Georgia Association of Regional Commissions

i = year, j = county.

percentage of the population with a bachelor’s degree or higher The bers sum to 339, which, when compared to the sums for other counties forthe year, means that Appling County ranks 49th best in education Themethodology section of the paper explains the QOL variable construction ingreater detail.

num-GEORGIA PROPERTY TAX SYSTEM

Georgia’s property tax system is fairly typical Property tax is assessed on thevalue of residential real property; commercial, business, and farm real prop-erty; and personal property, such as automobiles The Board of TaxAssessors assesses property at the county level All property
including land,structures permanently attached, and equipment, machinery, and fixtures

is assessed at 40% of its FMV The sum of three property tax rates
school,county, and state
constitutes the state’s total property tax rate

In 2009, 61.49% of total property tax revenues were allocated to theschool tax, 33.65% to the county tax, and 0.85% to state property tax(Georgia Department of Revenue Property Tax Administration AnnualReport FY2010) The tax, or millage, rate in each county is set annually,after the Board of County Commissioners (or other governing authority ofthe taxing jurisdiction) and the Board of Education determine propertyassessment values.5

simi-lar to other states in its local government practice and reliance on propertytaxes, which suggests that the results should be relevant to other states.There are, however, some distinctive features In Georgia, county govern-ments conduct property tax assessments annually to determine if they are

at the appropriate levels This feature is important, noteAlm et al (2011),because the research design can make use of all the years of available data

If property tax assessment occurred biannually or even less frequently,fewer years of data could be incorporated

Georgia has very few limitations on property tax It is not necessary, forexample, to obtain taxpayer approval for rate changes, and there are nolimits on general assessment, although in 2009, after our sample ended, astatewide temporary freeze on assessments was imposed.6Also, legislationthat became effective January 1, 2000, established the “Taxpayer Bill ofRights.” One of whose main thrusts was the prevention of “back-door taxincreases,” or indirect property tax hikes on properties that increased invalue because of inflation The state’s Department of Revenue adopted

Revenue Rule 560
11
2
.58 to roll back the millage rate when the taxdigest value increased because of reassessments.7The rule became effective

on November 14, 2000.8

These features matter empirically because the assessed property valuesand property tax rates are subject to variation every year That Georgiahas 159 counties benefits this study because of the large sample size andthus power of tests; the disclosure environment provides an extensiveamount of data Finally, Georgia is not an outlier on such measures aspopulation (9th) or square miles (23rd).9

property values, average county millage rates, and total property tax ues for each year of our sample, 1999
2009 From 1999 to 2008, assessedproperty values increased between 6% and 9% annually, but had only aslight increase from 2008 to 2009 because of the housing recession.10 Intotal, assessed property values increased from $187 billion in 1999 to

reven-$383.8 billion in 2009 Meanwhile, average property tax rates onlyincreased from 24.35% in 1999 to 26.27% in 2009 The average millagerate actually decreased in three of those years, 2000, 2006, and 2007

Table 2 Property Values and Property Taxes

Total Property Tax Revenue

Total Assessed Values of Residential Property

a The reports’ values are assessed as FMV × 40% assessment ratio.

b Millage rates are tax per $1,000 of property value.

Residential and commercial property taxes provide the two largestsources of Georgia property tax revenues In 2009, for example, tax reven-ues from assessed values of residential and commercial property totaled

$2.05 billion and $1.04 billion, respectively Industrial property tax tions added $220 million Other types of property taxes, such as agricul-ture, public utilities, mobile homes, timber, and heavy-duty equipmenttaxes tend to be much lower in comparison to residential and commercialproperty taxes, while motor vehicle taxes are only slightly higher thanindustrial property taxes

collec-The Georgia data used in this study fit primarily in the aggregate gory described byGuilfoyle and Rutherford (2000) They explain that capi-talization studies can be divided into three broad categories
aggregate,micro, and natural experiments
that exploit a policy or other settingchange Aggregated house prices and tax figures (e.g., median house priceand community tax rate) typify that category Micro studies use individualhouses as observations A benefit of aggregated studies is that they contain

cate-a lcate-arge number of communities cate-and cate-a lcate-arge cate-amount of scate-ample vcate-aricate-ation intax rates, but the aggregated house value is of lower quality than individualhouse measures Micro studies have a higher-quality dependent variable,actual sales prices, but tend to involve fewer communities; thus, there isless tax rate variation A point not made in their review article is that saletransactions include only a small fraction of the housing market becausesuch a small percentage of houses sell each year

The actual data in the current study overcome some of the historicshortcomings of aggregate studies The gross assessed values reflect all resi-dential property, similar to Wassmer (1993), as opposed to using a singleamount (like the median) to represent all properties Further, taxes aremeasured using all taxes paid, not just a mechanical calculation using statu-tory rates and assessment ratios

SAMPLE, DATA, AND MODEL SPECIFICATION

Sample

We analyze Georgia county data from 1999 to 2009 in order to assesswhether residential property values are associated with county effectiveproperty tax rates Further, we examine the relations between residentialproperty values and the four QOL variables for each county We chose theperiod from 1999 to 2009 due to data availability and state restrictions

Some of our variables are not available prior to this period, resulting in a

1999 start date Alternatively, a statewide temporary freeze on assessmentswas imposed in 2009 However, it didn’t become effective until school year

2010 after our sample ended We collect data at the county level and lyze data for all 159 counties in Georgia

ana-Data

We hand-collected from The Georgia County Guide most of the data for theindicators used to derive the QOL factors and for many of the remainingindependent variables.11 The Georgia County Guide is published by theCenter for Agribusiness and Economic Development at the University ofGeorgia All the data are compiled from sources available publicly fromvarious agencies, including the Georgia Department of Labor, GeorgiaDepartment of Revenue, Georgia Department of Education, and the U.S.Census Bureau, among others Almost all data are compiled on a year-by-year, county-by-county basis However, some variables, such as the U.S.Census data, are collected less frequently We hand-collected data from thewebsite Partner Up! for Public Health to measure the variables LifeExpectancy and Percentage of Obese Adults in a county We usedMapQuest to measure the distance from the county seats to Atlanta andthe U.S Bureau of Labor Statistics to measure consumer price index (CPI).Finally, the Region variable is obtained from the Georgia Association ofRegional Commissions Table 1 includes a list of the variables and theirdescriptions and data sources

The Georgia County Guide is missing a few years of data for the tor variables in this study We were able to collect some missing data fromthe original sources For other data we made the following adjustments:year 2000 data are used in place of year 1999 and 2001 data for the factorused to measure the percentage of the population age birth to 18; year 2000data are used for the year 2001 data for the factor used to measure the per-centage of the population 65 or older; and year 2002 data is used for year

indica-2003 data for the factor that measures average weekly wage

Model Specification

We estimate the following regression model using OLS regression12 toexamine the relation between residential property values and county effec-tive property tax rates and QOL factors:

RESVALUEij¼ B0þ B1ETRijþ B2ECrnkijþ B3EDrnkijþ B4HCrnkij

þ B5PSrnkijþ B6SALESTXijþ B7lATLjþ B8lPOPij

þ B9RURALjþ B10AGE65ijþ B11AGE018ijþ B12INDDISTj

þ B13CPIþ B14BUSVALUEijþXr11j¼r1BjREGIONjþ eij

ð1Þwhere

RESVALUEij= the assessed value of property in a county, grossed up

by the 0.4 statutory assessment ratio, to approximate FMV;

ETRij = the effective property tax rate for each county, calculated astotal local property taxes paid/(assessed value of property/0.4);

ECrnkij= annual ranking of the strength of the economy in a county;EDrnkij= annual ranking of education in a county;

HCrnkij= annual ranking of the health of a county;

PSrnkij= annual ranking of the public safety of a county;

SALESTXij= sales tax rate for each county;

lATLj = log of the number of miles from the county seat to theAtlanta airport;

lPOPij= log of the population of each county;

RURALj= 1 if a county is classified as rural and 0 if the county is sified as urban or suburban;

clas-AGE65ij= percentage of the population 65 or older;

AGE018ij= percentage of the population age birth to age 18;

INDDISTj= 1 for the following counties: Fulton, Haralson, Gwinnett,Gordon, Carroll, Bartow, Walker, Jackson, Whitfield, Dekalb, Laurens,Hall, Cobb, Mitchell, Floyd, Walton, Thomas, Chattooga, Toombs, andLowndes; and 0 otherwise;

CPIi= annual average consumer price index;

BUSVALUEij = the per capita assessed value of commercial property

in a county, calculated as the value of business property/population inthe county;

REGIONj= a dummy variable for each region, 1 to 11

The subscript i represents a year, while the subscript j represents a county

Dependent Variable: RESVALUE

The dependent variable RESVALUE, or property value, is measured usinggross digest assessed value, grossed up by the 0.4 statutory assessment ratio

to approximate a FMV, similar toWassmer (1993) The log transformation

is the most widely used functional form in the incidence literature

actual home sale transaction data or assessed value to measure marketvalue, the ultimate variable of interest Both proxy for an unobservablemarket value and have been used in prior research (Sirmans, Diskin, &

there could be relatively few sale transactions or the sales that occur arenot representative of the overall market

For practical data collection, the sale price approach necessitates arestricted geographical area in order to produce a data set with home-specific details like square footage, age, etc while still being able to assume

“all else equal.” Data availability and access are also issues since there is

no electronic database containing all sale transactions A benefit of usingassessed value is that all property is included in the measure and tax exemp-tions are taken into consideration On the other hand, valuation assessment

is difficult and can result in non-uniformity in property tax administration

Effective Tax Rate: ETR

ETR, or the effective tax rate, is the most popular and theoretically priate measure (Sirmans et al., 2008) of tax rate, and is based on theamount of taxes paid.13 Other studies have measured taxes as total taxespaid or the nominal tax rate A true measure of ETR is tax paid divided byproperty value Consistent with Wassmer (1993), our tax measurement istotal taxes paid (which incorporates property tax exemptions) divided byassessed property grossed up to FMV, where the gross assessed amount isbefore property tax exemptions

appro-QOL Variables

There are four QOL variables in the model: ECrnk, EDrnk, HCrnk, andPSrnk Each comprises multiple indicator variables that we rank individu-ally; the sum of those rankings forms an overall composite score on each ofthe four measures, economy, education, health and public safety, of eachcounty Using the composite scores, we rank the counties again to deter-mine the overall ranking on each of the QOL variables.14 For the QOLvariables, the county with the worst measurement is ranked 1 and the best

is ranked 159 If two counties have the same value, they both receive thehigher, or better, ranking; and the next-highest county is ranked two num-bers below Ranking the best county highest rather than No 1 improvesinterpretability of regression results We expect a positive association

between each of the four QOL variables and housing values In addition,

we ran a regression that included each of the individual variables thatmake up the four factors and footnote those results in the section thatincludes the multivariate results

The indicators for the economy variable, ECrnk, include income percapita, annual unemployment rate, poverty rate, and average weeklywage.15 All four of these indicators measure the health of a county’s econ-omy As income per capita and average weekly wage increase, so does thecounty’s QOL ranking Thus, the county with the highest value on each ofthese two indicators is ranked best, or 159 However, lower values ofannual unemployment rate and poverty rate are expected to increase acounty’s economic QOL ranking Therefore, the county with the lowestvalue on each of these two indicators is ranked best, or 159

The indicators for the education variable, EDrnk, include percentage ofthe population lacking basic literacy skills, high school dropout rate,teacher-student ratio, education funding per student, and percentage of thepopulation with a bachelor’s degree or higher Education is by far thelargest county expenditure and target for property tax dollars Becausehigher values of teacher-student ratio, education funding per student, andpercentage of the population with a bachelor’s degree or higher are gener-ally acknowledged to improve a county’s education environment, thecounty with the highest value on each of these variables is ranked best, or

159 Alternatively, lower levels of percentage of the population lackingbasic literacy skills and high school dropout rates improve the educationenvironment of a county Therefore, the county with the lowest level ofeach of these is ranked best, or 159

The indicator variables for the health factor, HCrnk, include life tancy, infant mortality rate, percentage of uninsured population, low birthweight (total rate per 100 live births), and percentage of obese adults Twoindicators, life expectancy and percentage of obese adults, are not collected

expec-on an annual basis because of data availability at the county level We usethe 2006 county-level value of life expectancy and the 2007 county-levelvalue of percentage of obese adults for each year of the study for eachcounty The county ranking highest on life expectancy is ranked best, or

159, whereas counties with the lowest infant mortality rates, percentage ofuninsured population, low birth weights, and percentage obese adults areranked best, or 159

The indicators for the public safety variable, PSrnk, include violentcrimes reported per capita (we collect murder, rape, robbery, and aggra-vated assaults); property crimes reported per capita (burglary, larceny, and

motor vehicle thefts); juvenile arrests per capita; and adult arrests percapita Public safety is the second-highest expenditure of property tax col-lections, including expenditures for police, county jails, and courthouses.The county with the lowest ranking on each of these four factors is rankedthe best, or 159.

Control Variables

We include a sales tax rate variable, SALESTX, for each county as a trol variable Sales taxes are of interest to this study because they generatetax revenues that could affect the county’s property tax environment Priorresearch often finds that taxes suppress growth and economic development,which could affect housing values.16Jung (2001) shows that for the period1984
1997, Georgia counties that added a local option sales tax experi-enced property tax relief Thus, a predicted relation is not clear; if propertytaxes and sales taxes are substitutes, a negative relation is expected betweenSALETX and ETR However, if a jurisdiction tends toward a high or lowoverall tax environment, a positive relation could result.17

con-This study also includes a number of variables to control for factorsother than taxes that might be associated with the county housing values.Prior property tax research has employed the control variables we use.Some variables are included to capture basic demographics, while othersare intended to capture the greater service needs and related expenditures

of certain segments of the population Consistent withDye, McGuire, and

the county seat to the Atlanta airport, to control for the distance to thecentral business district This variable is measured using MapQuest Weexpect house values to be higher the closer a county is to Atlanta

Two variables
lPOP, the log of the population of each county, andRURAL, coded 1 if the county is classified as rural or 0 if classified asurban or suburban
control for the effect of the size and dispersion ofpopulation on housing values.18The denser population and greater scarcity

of land in urban or suburban areas could create a more inelastic housingsupply and result in a greater capitalization of taxes and services into hous-ing prices in urban/suburban areas Alternatively, if new or changing hous-ing options do not exist or if rural areas are landlocked due to large landownership by residents this could produce a more inelastic housing supply

in rural areas resulting in a greater capitalization of taxes and services We

do not predict directions for lPOP and RURAL

We include AGE65, the percentage of the population age 65 or older,and AGE018, the percentage of the population age birth to age 18, to

control for possible differences in housing needs, locations, and public vices and benefits that a county might emphasize if one age group or theother is highly represented Georgia’s school district structure consists of

ser-159 county-based school districts and 22 municipal or “independent”school districts We include a control variable, INDDIST that equals 1 if acounty has one or more independent school districts within its borders Wealso include the CPI and region (REGION)19dummy variables to controlfor any cost-of-living or geography-related systematic effects on residentialproperty values.20Finally, we include BUSVALUE, the per capita assessedvalue of commercial property in a county, to control for the effects of busi-ness development on property valuation Again,Table 1 lists the variablesand their descriptions and data sources

RESULTSDescriptive Statistics

logged residential property values, RESVALUE, are 20.22 and range from16.63 to 25.07 The mean ETR for residential property is 0.011 The rangefor this measure is 0.003
0.02, indicating variance among the counties.The four QOL variables are the sums of the county rankings from 1 to 159,

so the descriptive statistics are very similar The slight variations are due toties that occur in the rankings If two counties have the same value, theyboth receive the higher, or better, ranking; the next-highest county isranked two numbers below

County sales tax rates vary little The minimum county tax is 1.00% andthe maximum is 3.00%, with an average of 2.82% and a median of 3.00%.The age variables have considerable variation as well The percentage ofthe population age 65 or older, AGE65, ranges from 1.45% to 28.88%,with a mean of 12.26% The variable AGE018, the percentage of thecounty population age birth up to 18, ranges from 15.87% to 38.29%, with

a mean of 25.79% The variable CPI ranges from 162.00 to 208.68, with amean of 184.78 The per capita assessed value of commercial property in acounty, BUSVALUE, varies from 300.38 to 21,738.48, with a mean

of 3,721

vari-ables in our model The dependent variable, RESVALUE, has a cally significant negative correlation with ETR and one of the QOL

statisti-variables, HCrnk (health) The variables ECrnk (economy), EDrnk tion), and PSrnk (public safety) have a significantly positive correlationwith RESVALUE lPOP also correlates positively and significantly withRESVALUE, while RURAL has a significantly negative relation withRESVALUE, leading to an initial conclusion that more populated and lessrural counties have higher gross housing values Our correlation resultsindicate that relatively high proportions of older (AGE65) residents tend tolive in counties with lower gross housing values, while higher proportions

(educa-of younger residents (AGE018) live in counties with higher gross housingvalues RESVALUE is significantly correlated with all of the control vari-ables in our model

Multivariate Results

We present the results of our model inTable 5 Our variables of interest
ETR and three of the QOL variables, ECrnk, EDrnk, and PSrnk
havethe predicted relationships with RESVALUE, the FMV of housing in a

Table 3 Descriptive Statistics

Variable n Mean Median Std Dev Minimum Maximum RESVALUE 1,749 20.22 20.14 1.57 16.63 25.07 ETR 1,749 0.011 0.011 0.002 0.003 0.02 ECrnk 1,749 80.16 80.00 45.90 1.00 159.00 EDrnk 1,749 80.17 80.00 45.90 1.00 159.00 HCrnk 1,749 80.22 80.00 45.92 1.00 159.00 PSrnk 1,749 80.22 81.00 45.91 1.00 159.00 SALESTX 1,749 2.82 3.00 0.41 1.00 3.00 lATL 1,749 4.71 4.82 0.66 2.20 5.78 lPOP 1,749 10.11 9.99 1.15 7.50 13.85 RURAL 1,749 0.44 0.00 0.50 0.00 1.00 AGE65 1,749 12.26 12.36 3.37 1.45 28.88 AGE018 1,749 25.79 26.06 2.67 15.87 38.29 INDDIST 1,749 0.11 0.00 0.32 0.00 1.00 CPI 1,749 184.78 181.80 15.58 162.00 208.68 BUSVALUE 1,749 3721 3173.24 2553.18 300.38 21738.48 Note: See Table 1 for variable descriptions.

Table 4 Correlation Coefficients.

Variable 1 RESVALUE 2 ETR 3 ECrnk 4 EDrnk 5 HCrnk 6 PSrnk 7 SALESTX

county ETR has a strong negative association with RESVALUE (p-value<0.0001) Thus, higher effective property tax rates are associated with lowerhousing values, indicating that property taxes are capitalized into the price

of housing in Georgia counties (We estimate the rate of capitalization inthe next section.) Three of the four QOL variables have a statistically signifi-cant and positive relation with housing values, ECrnk (p-value < 0.0001),EDrnk (p-value = 0.0240), and PSrnk (p-value = 0.0006), while the remain-ing QOL variable HCrnk is positive but not significant (p-value = 0.1946)

Table 5 Regression Results from Model 1

RESVALUEij¼ B 0 þ B 1 ETRijþ B 2 ECrnkijþ B 3 EDrnkijþ B 4 HCrnkijþ B 5 PSrnkij

þ B 6 SALESTXijþ B 7 lATLjþ B 8 lPOPijþ B 9 RURALjþ B 10 AGE65ij

þ B 11 AGE018ijþ B 12 INDDISTjþ B 13 CPIiþ B 14 BUSVALUEij

þXr11j¼r1B j REGIONjþ e ij

Variable Parameter Estimate Std Error t-Statistics p-Value INTERCEPT 7.929 0.313 25.28 < 0.0001 ETR
49.720 3.781
13.15 < 0.0001 ECrnk 0.004 0.000 15.33 < 0.0001

BUSVALUE 0.00003 0.000 6.54 < 0.0001 REGION DUMMIES INCLUDED INCLUDED INCLUDED INCLUDED Adj R2= 0.9645.

Model F-Statistics = 1901.78 (p-value < 0.0001).

Note: See Table 1 for variable descriptions.

Therefore, higher county rankings on QOL factors are generally associatedwith higher gross county housing values.21 This result is consistent withQOL public services and benefits capitalized into the price of homes.22

We do not find a significant relation between county sales tax rates andhousing values in Georgia counties This could be because the sales taxrates have little variation They range from 1.00% to 3.00%, with a mean

of 2.82% and a median of 3.00%, indicating that more than half of thecounties have sales tax rates of 3.00%

The remaining control variables help us better understand the ship between county demographic variables and gross county housingvalues As expected, lATL, the log of the number of miles from the countyseat to the Atlanta airport, is negative and significant (p-value < 0.0001).This result suggests that counties farther away from Atlanta have lowerhousing values The log of population, 1POP, is positive and highly signifi-cant (p-value < 0.0001) Hence, housing prices are higher in more heavilypopulated counties The variable RURAL, coded 1 if the county is classi-fied as rural or 0 if classified as urban or suburban, is negative andsignificant (p-value < 0.0001), indicating that rural counties have lowerhousing prices

relation-We include our next two variables, AGE65, the percentage of the countypopulation age 65 or over, and AGE018, the percentage of the countypopulation age birth to 18, in the model to control for possible differences

in housing needs, locations, and public services and benefits that might beemphasized in a county if one age group or the other is highly represented.AGE65 is positive and significant (p-value < 0.0001), while AGE018 isnegative but not significant (p-value = 0.3631) The Pearson correlationcoefficient between AGE65 and RESVALUE is negative and significant,but after controlling for the economy of a county, healthcare, and othervariables, we find that a larger proportion of residents aged 65 or older isassociated with higher housing values INDDIST has a significantly nega-tive coefficient (p-value = 0.0004), suggesting that gross housing values arelower in counties with independent school districts CPI has a significantlypositive coefficient (p-value < 0.0001), thus confirming a positive relationbetween inflation and housing prices We also include the variableBUSVALUE to control for business property values in a county Its coeffi-cient is positive and significant (p < 0.0001), indicating that housing valuesare higher in counties with higher values of business property Lastly, weinclude region dummy variables to control for any location effects Therewere several significant regions We also ran our model deleting bothFulton and DeKalb counties to make sure that Atlanta is not driving the

results All variables were in the same direction and significant at the samelevels In another sensitivity test, we tested a composite QOL ranking Weaveraged each county’s rankings on the four QOL areas to come up withone composite score for each county The composite QOL variable is posi-tive and highly significant (p-value< 0.0001) All other variables remain inthe same direction and significant at the same levels as those in the origi-nal model.

Incidence of Tax: The Rate of CapitalizationPrior research on residential property tax capitalization and incidence teststhe relation between property values and property taxes, and if there is astatistically negative relation, the extent of capitalization (i.e., the portion

of the tax borne by property owners) can be estimated Measuring taxcapitalization depends on assumptions about discount rates and timehorizon, and can have large effects on interpreting capitalization results

studies calculate a point estimate, and some test a hypothesis that there isnot full capitalization.Sirmans et al (2008)review 28 studies and find vary-ing results, from no capitalization to overcapitalization

We followMan’s (1995)point estimate calculations to determine a rate

of capitalization and find rates that range from 101% to 147%, depending

on assumptions.23 These results suggest that past results for capitalizationrates could be underestimated because of measurement error in publicservice variables and the spurious correlations that are due to positive colli-nearity between tax rates and public service input measures (Palmon &

One reason for overcapitalization could be that property owners haveexpectations of additional assessments and/or increases in their nominalproperty tax rates Thus, expected nominal tax rate increases may also becapitalized and indicate overcapitalization of the existing rate It also could

be a function of different expectations, meaning that homeowners withover-assessed property anticipate no future decreases or owners withunder-assessed property anticipate future increases (Sirmans et al., 2008),and both will overcapitalize

Finally, we determine the elasticity of housing prices with respect to theETR by multiplying the coefficient on the effective property tax rate vari-able (ETR) by its mean value (Man, 1995) This calculation produces a

housing price elasticity of
0.547, which means that a 1.0% increase in theeffective property tax rate reduces county housing values by 0.547%.

Marginal Benefits of QOL FactorsPrior tax capitalization research has not fully examined the marginal bene-fits of public services While our regression results indicate that a bettereconomic environment and more favorable rankings on education, publicsafety, and health are associated with higher housing values, we also exam-ine each factor’s marginal effects We use t-tests of standardized coefficients

to determine which QOL factors have the greatest impact on ing values

hous-As expected, on a statewide basis, the economy (ECrnk) has the greatestinfluence However, none of the other three factors (education, health, andpublic safety) has a significant marginal benefit over the others Next wedivide the state into the 12 regions indicated by our REGION variable andfind that the marginal effects of QOL factors differ significantly by region.While in most regions, the economy variable affects housing pricesmore than the other factors, education is a close second In four regions,the economy (ECrnk) has a significant marginal benefit over education(EDrnk) in predicting housing values, while in three regions, education is asignificantly better indicator of housing values In the other five regions,neither the economy nor the education has a significant marginal benefitover the other in predicting housing values Finally, both the health(HCrnk) and the public safety (PSrnk) factors have a greater influence onhousing values than the economy factor in only one region each

CONCLUSIONThis study investigates the relation between residential property values andlocal property taxes and public services, both of which influence home-owner decisions about where to live and both of which are reflected in resi-dential property values Property tax capitalization theory predicts thatproperty values will be negatively related to a taxing jurisdiction’s propertytaxes and positively related to its public services Our paper is a first step inbroadly defining and quantifying public services and their marginal effects

on housing values by modeling local public services using primarily output

measures in four areas
economy, education, health, and public safety
that are intended to capture broad QOL aspects of local jurisdictions.

We find a strong negative relation between county ETRs and residentialproperty values This result is consistent with some portion of the residen-tial property taxes being borne by owners and capitalized into the price ofthe property We also examine QOL rankings in all four areas and findthat there is a significant positive association between three QOL measuresand residential property values in Georgia counties Thus, we find thatQOL measures are capitalized into property values, and that propertyvalues are partially determined by a county’s QOL and socioeco-nomic factors

Because a significant negative relation exists between ETRs and housingvalues, we calculate the incidence of the property tax or level of capitaliza-tion, and find that Georgia property taxes are capitalized into housingprices at a rate greater than 100%, based on our OLS results Therefore,property taxes in our sample are fully borne by owners of property, andmarket participants rationally discount properties subject to higher taxes.Accordingly, only unexpected tax changes can be passed on to new home-buyers while current owners are absorbing some portion of expected taxchanges Our finding of overcapitalization suggests that concerns of under-estimation of tax capitalization in prior research, which primarily finds par-tial capitalization, could be due to measurement error in public servicevariables and spurious correlations between tax rates and public serviceinput measures Overall, our results should encourage researchers in theproperty tax incidence area to fully consider public service benefits in theirtax capitalization models In addition, local government officials can alsobenefit from our evidence Our findings have implications for the competi-tive environment that local policymakers face in attracting residentsthrough wise tax and spending decisions on local public services

NOTES

1 The Oates model was criticized for being biased in that an increase in homevalue, because of increased public service, must be exactly offset by the increasedtax cost.Oates (1973)corrects for the earlier model deficiencies and finds full capita-lization of taxes

2 The review includes 2 business property studies and 26 residential erty studies

prop-3 See Easterly (1999); Kahn (2002); Becker, Philipson, and Soares (2003);

Mattey, Wascher, and Gabriel (2003); Shapiro (2005); Veenhoven and Hagerty

(2006); Easterlin and Angelescu (2007); Nyman et al (2007); Albouy (2008);

Granger and Gregory (2008); andRappaport (2009)

4 We recognize that QOL could be and has been measured in many ways.While most measures have significant similarities, arguments could be made forother measures of QOL We attempt to identify measures that cover a broad range

of factors that affect the lives of county residents and their property values, andthat local government officials could influence The Vocino study’s variables are amore comprehensive and better measure of overall QOL for the purposes of thisstudy than other prior research An example of a study that we do not follow forour QOL variable construction is Alzate (2005), who examined single mothers onwelfare in Georgia We considered the focus of her study too narrow for our prop-erty tax capitalization model It contains fewer factors and indicators; and a largerpercentage of the factors measure low income and poverty within a county

5 A tax rate of 1 mill represents a tax liability of one dollar per $1,000 ofassessed value For example, a house with a market value of $100,000 has anassessed value of $40,000 In a county where the millage rate is 25 mills, the prop-erty tax on that house would be $1,000 (i.e., $25 for every $1,000 of assessed value,

or $25 multiplied by 40)

6 Alm et al (2011)do not identify any other changes, but Senate Bill 55, passed

in 2009, added foreclosure and distressed sales in the section of the Official Code ofGeorgia that lists criteria to be used in determining FMV Dana Eaton, chiefappraiser for Troup County, said this change primarily affected residential prop-erty She did not identify any other important appraisal law changes during theperiod 1999
2009 Alm et al (2011)address the fiscal positions of local govern-ments, given that the recession had major effects on federal and state governments.They conclude that while there is state variation, local governments on averagehave not experienced similar large, negative budgetary effects Their analysesinclude the national level (19 states) and Georgia as a case study, with the aim toexplain factors that affect local source school revenues Like the current study,their case study is feasible due to the rich set of data available for Georgia localproperty taxes

7 The tax digest value is the dollar value of all assessments of real and tangiblepersonal property subject to taxation

8 This rule established the procedures for the computation of a rollback millagerate by levying and recommending authorities as a result of increases in the value ofexisting real property value due to inflation and the requirements of advertisingnotices of public hearings, press releases, and the authority’s intent to increase prop-erty taxes when the proposed millage rate exceeds the computed rollback rate(Georgia Department of Revenue, Local Government Services Division)

9 Sources: U.S Census Bureau for population (2010 data) andnetstate.comforsquare miles

10 All figures in this paragraph come from Georgia Department of RevenueProperty Tax Administration Annual Report FY2010

11 Current year data from The Georgia County Guide can be purchased

at https://estore.uga.edu/C27063_ustores/web/product_detail.jsp?PRODUCTID=4858&SINGLESTORE=true Interested parties can obtain prior years’ data free ofcharge using this same web address